116
Energy & Fuels 1997, 11, 116-125
Coal/Petroleum Residuum Interactions during Coprocessing under Noncatalytic, Low Solvent/Coal Ratio Conditions Jasna Tomic`† and Harold H. Schobert* Fuel Science Program, Department of Materials Science and Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802 Received June 26, 1996. Revised Manuscript Received October 17, 1996X
Coprocessing reactions were carried out using five coals and three petroleum residua at 350, 400, and 450 °C with N2 and H2. No catalyst was used. The interactions between the coal and the petroleum residua were investigated in terms of following the conversions and yields of THFinsolubles. The influence of temperature, time, petroleum residua, and coal characteristics was examined. Conversions at or below 400 °C are primarily determined by the nature of the coal. The characteristics of the residuum have little influence on the coal conversions. The residuum behaves like a liquid medium for the coal fragments at this temperature regime and does not have properties of a hydrogen donor or hydrogen shuttler. The maximum coal conversion is ≈40 wt %, comparable to the conversions obtained in pyrene under the given conditions. At temperatures >400 °C retrogressive reactions are observed, indicated by low or even overall negative conversions (-20 wt %). The retrogressive reactions are a result of the interactions between the coal and petroleum residuum fragments and are influenced by the properties of the coal as well as the petroleum residuum. Both the coal and the residuum must be undergoing thermal decomposition for the retrogressive reactions to occur. Residua that have a higher degree of polycondensation (Har/Car e 0.4) may be inferior coprocessing solvents at temperatures >400 °C. Similarly, noncaking coals proved to be less desirable than caking coals.
Introduction Coal/petroleum coprocessing is an alternative approach for production of distillable liquids relative to traditional coal liquefaction or hydroprocessing of petroleum residua. The use of lower cost petroleumderived, instead of coal-derived, solvents, is an advantage of coprocessing over traditional liquefaction. Coprocessing can be viewed as a process in which heavy petroleum feedstocks are used as a solvent or hydrogen donor in the liquefaction of coal, but it can be also viewed from another perspective as a process of hydrotreating the heavy petroleum feedstock in which coal is used as an additive. From one standpoint, coprocessing is more complex than liquefaction due to the different nature of the reactants. Coals are solids for which atomic H/C ratios are generally 40% were also found to interact more with coal.5,6 In addition to hydrogen-shuttling properties, aromatic (1) Moschopedis, S. E.; Hawkins, R. W.; Fryer, J. F.; Speight, J. G. Fuel 1980, 59, 647-653. (2) Curtis, C. W.; Guin, J. A.; Pass, M. C.; Tsai, K. J. Fuel Sci. Technol. Int. 1987, 5, 245-275. (3) Rahimi, P. M.; Dawson, W. H.; Kelly, J. F. Fuel 1991, 70, 9599. (4) Flynn, T.; Kemp, W.; Steedman, W.; Bartle, K. D.; Burke, M.; Taylor, N.; Wallace, S. Fuel Process. Technol. 1989, 23, 197-204. (5) Wallace, S.; Bartle, K. D.; Burke, M. P.; Egia, B.; Lu, S.; Taylor, N.; Flunn, T.; Kemp, W.; Steedman, W. Fuel 1989, 68, 961-967. (6) Bartle, K. D.; Bottrell, S.; Burke, M. P.; Jones, C.; Louie, P. K.; Lu, S.-L.; Salvado, J.; Taylor, N.; Wallace, S. Int. J. Energy Res. 1994, 18, 309-315.
© 1997 American Chemical Society
Coal/Petroleum Residuum Interactions
Energy & Fuels, Vol. 11, No. 1, 1997 117 Table 1. Analysis of Project Coals coal
seam state ASTM rank moisture,a wt % ash,b wt % % Cc %H %N %O % Stotal FSI T, °C max fluid a
PSOC 1488
PSOC 1498
PSOC 1501
PSOC 1504
PSOC 1448
Dietz Montana subB 23.7
Wadge Colorado hvCb 9.4
Juanita C Colorado hvBb 5.8
Upper Sunnyside Utah hvAb 3.4
York Canyon New Mexico hvAb 1.4
5.3 76.0 5.2 0.9 17.3 0.5 0.0 n/a
7.1 77.5 5.4 1.8 14.7 0.6 0.5 n/a
5.5 80.5 5.3 1.5 12.0 0.7 2.0 421
7.5 82.0 5.8 1.7 9.7 0.8 5.5 433
11.4 84.9 6.0 1.8 7.5 0.5 8.0 438
As received. b Dry basis. c Dry-ash-free basis.
compounds have a relatively good solvating ability toward coal fragments. Addition of an aromatic component, such as anthracene oil, to a coal/petroleum mixture can enhance the solvating capacity of petroleum feedstock.7 Conversely, several studies also report the beneficial effect of the addition of small amounts of coal during petroleum upgrading processes.8-10 The extent of coal conversion is influenced by the choice of reaction conditions. A review of the literature shows that different starting feedstock will have different optimum reaction conditions.11-14 Change in reaction temperature influences differently coal conversions depending on the petroleum feedstock used. Generally, temperatures in the range 300-475 °C have been used in coprocessing studies, and optimum reaction temperatures appear to be in the vicinity of 400 and 425 °C. As examples, Curtis et al.12 report maximum oil conversion and minimum insoluble matter at 425 °C, and Font et al.15 report a coal conversion peak at 417 °C. However, a temperature rise above the optimum is accompanied by retrogressive reactions that are observed in the higher gas and insoluble residue yields. Retrogressive reactions may occur at higher temperatures as well as at relatively longer reaction times. Prevailing values of reaction times used in laboratory coprocessing investigations are in the range of 15-90 min.14-18 The optimum reaction time depends on the specific feedstock as well as the presence of a catalyst. Although a catalyst contributes to the increase of overall coal conversions, the effects of the different feedstock (7) Rincon, J. M.; Ramirez, J.; Cruz, S. Fuel 1990, 69, 1052-1054. (8) Fouda, S. A.; Kelly, J. F.; Rahimi, P. M. Energy Fuels 1989, 3, 154-160. (9) Miyake, M.; Takahashi, K.; Kigashine, J.; Nomura, M. Fuel Process. Technol. 1992, 30, 205-213. (10) Yoshida, T.; Nagaishi, H.; Sasaki, M.; Yamamoto, M.; Kotanigawa, T.; Sasaki, A.; Idogawa, K.; Fukuda, T.; Yoshida, R.; Maekawa, Y. Energy Fuels 1995, 9, 685-690. (11) Moschopedis, S. E.; Hawkins, R. W.; Speight, J. G. Fuel Process. Technol. 1982, 5, 213-228. (12) Curtis, C. W.; Tsai, K. J.; Guin, J. A. Ind. Eng. Chem. Process Des. Dev. 1985, 24, 1259-1266. (13) Curtis, C. W.; Tsai, K.-J.; Guin, J. A. Ind. Eng. Chem. Res. 1987, 26, 12-18. (14) Ceylan, K.; Stock, L. M. Energy Fuels 1991, 5, 482-487. (15) Font, J.; Fabregat, A.; Salvado, J.; Moros, A.; Bengoa, C.; Giralt, F. Fuel 1992, 71, 1169-1175. (16) Miller, T. J.; Panvelker, S. V.; Wender, I.; Tierny, J. W. Fuel Process. Technol. 1989, 23-38. (17) Rosal, R.; Cabo, L. F.; Diez, F. V.; Sastre, H. Fuel Process. Technol. 1992, 31, 209-220. (18) Fabregat, A.; Salvado, J.; Giralt, J.; Moros, A.; Font, J.; Giralt, F. Int. J. Energy Res. 1994, 18, 317-328.
and the coal/petroleum interactions are masked by catalytic processes.19,20 Our goal was to investigate the interactions between the coal and the petroleum residua during coprocessing and to study the effect of coal type, petroleum residua type, temperature, and reaction time. For this reason noncatalytic reaction conditions were chosen as well as relatively low severity reaction conditions. This is an extension of our previous work in which we investigated the effect of different model compounds on coal conversion.21 In the present paper we report the results of coprocessing reactions between a series of coals and heavy petroleum residua and the evaluation of the effectiveness of the petroleum residua as process solvents determined on the basis of the extent of coal conversion and yield of THF-insolubles. Retrogressive reactions were intentionally not prevented as we specifically wanted to examine the extent of these reactions under different reaction conditions and for different coal/ residuum pairs. Experimental Section Materials and Characterization Techniques. Five coal samples used in this work were selected from the Penn State Coal Sample Bank and Data Base. The origins and analyses of the coals are given in Table 1. Prior to reaction, the coal samples were ground to 150 µm and dried to 1% moisture content. Three petroleum residua samples were used: Hondo, Blend vacuum residuum, and West Texas vacuum residuum. A vacuum residuum is usually the component of heavy oil emerging at the bottom of vacuum distillation column with a cutoff point around 525 °C.22 Hondo residuum was obtained from Unocal, and Blend and West Texas residua were obtained from Amoco. Hondo and West Texas samples are derived from their parent crudes, while Blend residuum is derived from a blend of eight different crudes. The properties reported in Table 2 were obtained by us, unless otherwise stated. These values differ from those reported previously,23 which were provided by the suppliers. The petroleum residua were used as received. (19) Curtis, C. W.; Tsai, K.; Guin, J. A. Fuel Process. Technol. 1987, 16, 71-87. (20) Ettinger, M. D.; Stock, L. M.; Gatsis, J. G. Energy Fuels 1994, 8, 960-971. (21) Tomic, J.; Schobert, H. H. Energy Fuels 1996, 10, 709-717. (22) Speight, J. G. The Chemistry and Technology of Petroleum, 2d ed.; Dekker: New York, 1991; 760 pp. (23) Tomic, J.; Schobert, H. H. Fuel. Process. Technol. 1993, 34, 295312.
118 Energy & Fuels, Vol. 11, No. 1, 1997
Tomic` and Schobert
Table 2. Properties of the Petroleum Residua wt
%a
asphaltenes maltenes saturates aromatics polars C H N S O API gravityb Vb (ppm) Nib (ppm)
Hondo Blend FHC-571 West Texas FHC-470 19.2 80.8 26.3 37.5 17.0 80.5 10.9 0.8 5.5 1.2 13.4 229 92
22.4 77.6 12.8 36.5 28.3 84.8 9.9 0.6 4.7 1.4 6.9 228 68
26.5 73.5 18.8 36.1 17.6 85.5 10.0 0.8 4.3 0.9 nd 13 18
a Units are wt % unless otherwise noted. b Provided by the suppplier.
To obtain more information regarding the composition of the residua, they were separated into solubility classes and fractions. First, they were fractionated into asphaltene (toluenesoluble/pentane-insoluble) and maltene (pentane-soluble) fractions using a modified procedure described by Eser.24 Approximately 10 g of sample was mixed with a small amount of toluene, which was added as a “thinner” to facilitate subsequent mixing of the toluene-solubles with pentane, and warmed in a water bath at 70 °C. An excess of pentane (400 mL) was placed into a separate beaker and vigorously stirred using a magnetic stirrer. The residuum was then added slowly to the excess pentane and stirred for 30 min. The mixture was left overnight to allow asphaltene precipitation before filtration using a Millipore vacuum filtration apparatus and a 0.45 µm pore size filter. The asphaltenes were dried in vacuum at 60 °C before being weighed. The amount of maltenes was determined by difference. The pentane-soluble fractions (maltenes) were further separated into compound classes following a procedure similar to that of Wallace et al.5 The sequential procedure provides a general idea of relative amounts and types of compounds in the pentane-solubles and, therefore, in the whole petroleum residua. A 35 × 2 cm glass column was packed with silica gel (Machery Nagel-60) of size range 0.04-0.063 mm. Approximately 300 mg of sample was diluted with dichloromethane and impregnated on ≈5 g of silica, which was added on top of the packed column. The column was then eluted successively with 100 mL of pentane (for saturates), benzene (for aromatics), and tetrahydrofuran (for polars). The elemental analyses of the petroleum residua and their fractions for carbon, hydrogen, and nitrogen were performed on a Leco CHN-600 elemental analyzer. The analyses were performed in duplicate using about 0.07-0.10 g of sample. The total oxygen and sulfur were determined by Galbraith Laboratories, Inc. (Knoxville, TN). The oxygen analysis was performed by pyrolysis using a Leco RO-478 analyzer, and sulfur analysis was performed by combustion on a Leco SC432 analyzer. Nuclear magnetic resonance (NMR) spectroscopy was used to obtain information on the chemical structures of the petroleum residua. 1H NMR spectra were obtained on a Bruker WM-360 spectrometer operating at a frequency of 360 MHz using a 90° pulse flip angle and 20 s relaxation delay time. Approximately 50 mg of sample was dissolved in 1 mL of deuterated chloroform. The different proton types were defined by chemical shift ranges relative to tetramethylsilane (internal standard) and chemical shift assignments given by Delppuech.25 13C NMR spectra were obtained at an operating (24) Eser, S. Ph.D. Dissertation, The Pennsylvania State University, 1986. (25) Delppuech, J. J. In Magnetic Resonance: An Introduction, Advanced Topics and Applications to Fossil Energy; Petrakis, L., Fraissard, J. P., Eds.; D. Reidel Publishing: Dordrecht, The Netherlands, 1984.
Table 3. Hydrogen and Carbon Types in the Petroleum Residua As Determined by 1H and 13C NMR %
Hondo
Blend FHC-571
West Texas FHC-470
Hara Holb ΗRc Ηβd Ηγe Carf CBg CRh Cβi CMe-Rj CEk CMe-γ Har/Car
10.7 2.2 7.9 52.7 26.5 25.1 30.2 15.9 15.0 6.9 1.5 5.4 0.67
10.5 0.8 12.7 60.7 16.2 28.4 16.9 16.8 23.2 8.2 1.4 5.1 0.51
10.0 0.1 9.9 61.4 18.6 33.1 23.9 13.0 15.0 6.5 2.1 6.4 0.42
a H , aromatic (6.3-9.3 ppm). b H , olefinic (4.5-6.3 ppm). c H , ar ol R methylene R to aromatic ring (2.1-4.5 ppm). d Hβ, paraffinic methylene or methyl (1.03-2.1 ppm). e Hγ, gamma or terminal methyl (0.05-1.03 ppm). f Car, aromatic (118-150 ppm). g CB, methylene bridge (37-60). h CR, CH alpha to aromatic ring other than CH3 (30-37 ppm). i Cβ, C beta to aromatic ring (23-30 ppm). j C k Me-R, CH3 alpha to aromatic ring (19-23 ppm). CE, CH3 gamma or further from aromatic ring (11-17 ppm).
frequency of 90.5 MHz. The spectra were obtained with inverse gate proton decoupling during acquisition, a pulse flip angle of 45°, and a relaxation delay of 10 s. Typically, 100 mg of sample was dissolved in 1 mL of deuterated chloroform and 1 wt % of relaxation agent, Cr (AcAc)3. The chemical shift assignment of carbon types were taken from Snape.26 The structural information of the three samples derived from the NMR analyses is presented in Table 3. Reaction Procedures. The reactions were performed in vertical microautoclave reactors (tubing bombs) of nominal 22 mL capacity which were heated to the desired temperature in a preheated fluidized sand bath. More detail on the reaction procedure is presented elsewhere.21,23 All reactions were conducted with nominally 2.5 g of coal and 5 g of petroleum residua at 350, 400, and 450 °C. The standard reaction time used was 30 min; selected experiments were carried out at 15, 45, and 60 min. In each case the reactors were pressurized to 3.5 MPa (at room temperature) with nitrogen or hydrogen. To determine the amount of THF-insoluble material from individually reacted coal and petroleum residua, each component was tested independently under the same reaction conditions. In this case the reactor was charged with 2.5 g of coal or 5 g of residuum. After the elapse of the desired time, the reactor was cooled by immersion in cold water. The gases were vented and the content of the reactor was washed out with tetrahydrofuran (THF) and separated into THF-insoluble material and THFsoluble material. The insoluble fraction was further extracted with additional THF in a conventional Soxhlet apparatus for 24 h. Excess THF was removed by rotary evaporation and the THF-soluble and THF-insoluble materials were dried in a vacuum oven at 110 °C for 12 h before being weighed. The final weight of the THF-insoluble matter was used to determine the coal conversion on a dry, ash-free (d.a.f.) basis. Duplicate and triplicate experiments indicate that the uncertainty of the coal conversion values is (2.5 wt %.
Results and Discussion Thermal Reactions of Petroleum Residua. The residua were reacted independently, without coal present, to investigate the influence of temperature. The yields of THF-insolubles under N2 and H2 are reported in Table 4. With one exception, the yields at 350 and 400 °C are relatively low, around 0.1 wt %. Overall, the (26) Snape, C. E. Anal. Chem. 1979, 51, 2189-2198.
Coal/Petroleum Residuum Interactions
Energy & Fuels, Vol. 11, No. 1, 1997 119
Table 4. Yield of THF-Insolubles (Weight Percent) from Petroleum Residua Reacted Independently at Three Temperatures under N2 and H2 (t ) 30 min) N2
H2
residuum
350 °C
400 °C
450 °C
350 °C
400 °C
450 °C
Hondo Blend West Texas
0.38 0.06 0.07
0.08 0.10 0.01
5.77 3.72 17.00
0.12 0.10 0.08
0.05 0.12 0.15
4.53 2.63 13.83
amount of THF-insoluble matter from thermal reaction of these petroleum residua by themselves is not significant at 350 and 400 °C. Jackson et al. have similarly shown, for different residua, that heating residua at 405 °C produced no detectable amounts of solids.27 On the other hand, at 450 °C the amounts of THFinsoluble solids from the petroleum residua are higher (2.6-17 wt %). Similar formation of substantial amounts of coke from residua at temperatures >400 °C has been reported.17 This behavior represents a distinct difference between the results reported in this paper and those we described from model compound studies.21 In that work, the yields of THF-insolubles from model compound blanks were e0.2% and were considered negligible. For all three residua, the yield of THF-insolubles at 450 °C is somewhat higher under nitrogen than under hydrogen, which was not apparent at the two lower temperatures. Hydrogen reduces the amount of insolubles from the residua samples. Similar results have been reported in upgrading of Maya residuum;28 H2 also increases coal conversion for coal “blank” reactions (i.e., coal reacted without residua present).21 Hydrogen pressure of 10-17 MPa is similarly used in industrial units to provide excess hydrogen to reduce coke formation and enhance production of lower boiling products.22 In our experiments the hydrogen pressure was 3.4 MPa; higher pressure would likely decrease the yield of insoluble matter even more. There is no clear distinction between the three residua in terms of the yield of THF-insoluble matter at the lower temperatures (350 and 400 °C). However, at 450 °C it is noticeable that West Texas residuum produces more THF-insoluble matter than either Hondo or Blend residuum (e.g. 2.6 vs 13.8 wt %). West Texas residuum has the highest asphaltene content of the three residua (Table 2). Other work, comparing the production of insolubles from two residua at 425 and 450 °C, showed that the higher amount of insolubles came from the one having the higher asphaltene content.17 Cracking of the residua becomes substantial as the temperature is raised to 450 °C. Cracking is a free radical process, strongly dependent on reaction temperature.22 During cracking the various hydrocarbon compound types present in residua will experience different transformations. Nonaromatic hydrocarbons and aromatic alkyl side chains will crack at elevated temperatures, leading to smaller molecules, while aromatic species will undergo condensation reactions.29 As the temperature increases further, condensation of aromatic compounds (27) Eamsiri, A.; Larkins, F. P.; Jackson, W. R. Fuel Process. Technol. 1991, 27, 149. (28) Heck, R. H.; Rankel, L. A.; DiGuiseppi, F. T. Fuel Process. Technol. 1992, 30, 69-81. (29) Fitzer, E.; Mueller, W.; Schaefer, W. In Chemistry and Physics of Carbon; Walker, P. L., Ed.; Dekker: New York, 1971; Vol. 7, pp 237-383.
becomes increasingly important, producing polycyclic aromatic systems. Polycyclic aromatics are identified as coke precursors and during petroleum processing can cause the buildup of large cokelike structures.30 Polycyclic aromatics31 or highly aromatic heavy oils27 have been identified as leading to coke formation. A significant proportion of polycyclic aromatics is present in asphaltenes, which may contain aromatic ring clusters up to seven rings with aliphatic chains from 1 to 20 carbons.32-34 In addition, asphaltenes are rich in heteroatoms, which also contribute to reactivity during cracking and hydrocracking, leading to the buildup of larger aromatic structures and the formation of solids. A recent paper by Kubo et al. shows that addition of small amounts (3-5 wt % of feed) of highly aromatic heavy fractions during heavy oil hydroprocessing decreases coke formation, probably due to removal of steric hindrance which then suppresses the hydrogenation of aromatic rings.35 Coking of the petroleum feedstock depends on its starting composition. Hondo residuum has the greatest aliphatic character of the three samples. This is evidenced by the low asphaltene content as well as the highest amount of saturates. The structural information from NMR (Table 3) further supports this view. Hondo residuum has the lowest carbon aromaticity (% Car ) 25), the highest proportion of hydrogen in paraffinic terminal methyl groups (% Hγ ) 26), and the highest proportion of carbon in alkyl side chains (% CB ) 30). The estimation of the polyaromatic ring sizes36 also indicates the smaller size in Hondo residuum (Har/ Car ) 0.67) than in the other two samples. Others have reported similarly that Hondo residuum may not be abundant in condensed structures.37 Blend residuum has a more aromatic character than Hondo (Tables 2 and 3). The fraction of saturates is low (13 wt %), while the proportion of polars is high (28 wt %). The size of the polyaromatic structures is also higher than in Hondo (Har/Car ) 0.51 vs 0.67 for Hondo). This suggests that Blend residuum is rich in aromatic ring structures substituted by shorter alkyl groups rather than longer, as was the case for Hondo residuum. Finally, West Texas residuum is the most aromatic. One-third of its carbon is aromatic (% Car ) 33) and is present in structures consisting of large polycondensed aromatic rings (Har/Car ) 0.42). The asphaltene content is also highest for West Texas residuum. From the various structural parameters and solvent fractionation studies, a clear differentiation can be established between Hondo residuum on the one hand and Blend and West Texas on the other. The differences between Blend and West Texas are less pronounced. However, the yield of THF-insoluble matter from the residuum reactions at 450 °C does not strictly follow this (30) Sullivan, R. F.; Boduszynski, M. M.; Fetzer, J. C. Energy Fuels 1989, 3, 603-612. (31) Chakma, A. Fuel Process. Technol. 1993, 36, 146-153. (32) Speight, J. G. Prepr.sAm. Chem. Soc., Div. Pet. Chem 1989, 34, 321-328. (33) Payzant, J. D.; Lown, E. M.; Strautz, O. P. Energy Fuels 1991, 5, 445-453. (34) Calemma, V.; Iwanski, P.; Nali, M.; Scotti, R.; Montanari, L. Energy Fuels 1995, 9, 225-230. (35) Kubo, J.; Higashi, H.; Ohmoto, Y.; Arao, H. Energy Fuels 1996, 10, 474-481. (36) For relative comparison, the Har/Car for pyrene is 0.62 and for tetracene is 0.67. (37) Trauth, M. D.; Yasar, M.; Neurock, M.; Nigam, A.; Klein, M. T.; Kukes, S. G. Fuel Sci. Technol. Int. 1992, 10, 1161-1179.
120 Energy & Fuels, Vol. 11, No. 1, 1997
Tomic` and Schobert
Table 5. Coal Conversion (Weight Percent daf) for Coprocessing and Coal “Blank” Reactions N2 residuum
350 °C
none Hondo Blend West Texas
3.6 17.7 18.7 15.6
none Hondo Blend West Texas none Hondo Blend West Texas
400 °C
H2 450 °C
350 °C
400 °C
450 °C
PSOC 1488 14.1 11.1 30.1 -6.0 29.9 -4.4 30.6 -19.0
9.3 12.3 18.3 20.9
19.5 28.7 33.1 34.6
22.9 0.5 -8.5 -15.4
3.4 12.5 12.3 10.6
PSOC 1498 15.9 11.8 24.1 -10.0 26.6 -8.3 25.5 -18.1
7.8 8.2 12.4 17.7
17.9 24.4 30.0 28.5
20.3 -0.8 -9.3 -18.6
6.0 15.1 13.0 16.0
PSOC 1501 22.2 16.9 31.5 1.3 31.1 -9.0 31.3 -9.7
7.4 13.8 15.4 22.0
25.3 30.5 33.5 32.6
18.9 3.2 -3.3
none Hondo Blend West Texas
7.5 19.8 14.8 14.4
PSOC 1504 21.5 17.3 39.4 3.0 34.4 -6.0 36.4 -10.5
10.4 18.6 17.6 15.8
22.7 33.7 36.5 34.2
21.2 8.6 3.7 -10.7
none Hondo Blend West Texas
10.4 18.7 18.3 19.6
PSOC 1448 23.8 22.7 35.8 1.3 38.0 -10.1 36.5 -15.1
15.5 18.2 15.8 17.1
27.9 40.8 36.8 40.7
25.2 5.7 0.5 -6.9
classification. In terms of the yield of solids, Hondo and Blend residuum seem more similar, while West Texas residuum stands on its own. Blend and West Texas are more aromatic and, thus, condensation and polymerization are the more likely processes occurring during cracking of these residua, leading to higher yields of solids. What distinguishes West Texas residuum is the presence of polycyclic aromatic structures of larger size, as determined by Har/Car, which provide its greater propensity toward coking and, ultimately, the observed higher solid yields. Effect of Temperature on Coprocessing Reactions. Coprocessing reactions were performed for all 15 coal/residuum combinations at three temperatures with both N2 and H2. The coal conversion values for these reactions are listed in Table 5. For comparison, the coal conversions of coal “blank” reactions are included. The coal blank results were discussed previously.21 Conversions of the coprocessing reactions are relatively low, which is not surprising as these reactions are without a catalyst and at relatively low pressures (3.5 MPa). At 350 °C the maximum coal conversion is 22 wt %, and at 400 °C conversions range from 24 to 41 wt %. These values are similar to those obtained by other investigators, who have reported noncatalytic conversions at 400 °C in the range of ≈30-35%.27,38 Conversions increased as the temperature was changed from 350 to 400 °C. The subsequent change in temperature to 450 °C produced a significant decrease in the observed coal conversion. In many cases the coal conversions at this highest temperature are negative, indicating that more solid matter is produced during the reaction than the initial weight of coal loading. Of the three temperatures we examined, 400 °C gives the highest conversions. This agrees with our earlier find(38) Curtis, C. W.; Hwang, J. Fuel Process. Technol. 1992, 47-67.
ing using model compounds21 and with reports in the literature,12,39,40 as well as with many papers showing onset of retrogressive reactions in the 400-450 °C range.17,18,22,41-45 In part this derives from a shift in equilibrium at high temperature (i.e., 450 °C) to favor dehydrogenation rather than hydrogenation.46 The insolubles can originate from the coal or from the petroleum residuum; thus, it has been a practice to correct the observed conversion for the amount of solids originating from the residuum.19,38 In a previous paper23 we corrected the conversion for the amount of THF-insolubles expected to originate from the residua but have abandoned this approach since the assumption that the residuum would behave (i.e, convert to THFinsolubles) in the same manner in a reaction with or without coal present may inadvertently oversimplify interpretation of results. Several studies have shown that small amounts of coal (up to 10 wt %) modify the behavior of the residuum during upgrading.8-10,23,47 Instead, the discussion below will compare the amount of THF-insolubles directly rather than “corrected” coal conversion values. The range of conversion values is narrow for the different coal/residuum pairs at 350 and 400 °C. Similar values were achieved regardless of the residuum used, and only a slight influence of the coal is noticeable. A similarly small influence of coal has been reported for the coprocessing of Chinese coals.48 Bolat and coworkers have shown that conversions of three Turkish lignites were very similar and attributed this to the fact that the lignites were of similar volatile matter content.45 Our coals also happen to have a very narrow range of volatile matter contentss41.5-45.2%sso that our findings of a relatively small influence of the specific coal selected agree with Bolat’s earlier work. At these temperatures there is no clear effect of the gas atmosphere. These results imply that the conversions are most strongly influenced by the reaction temperature, which affects thermal decomposition of the coal, as previously observed for the reactions with model compounds.21 The presence of the residuum increases coal conversion relative to coal “blank” experiments at these reaction temperatures but there is no clear trend as to one residuum having greater ability to improve coal conversions over another among this set of coalresiduum mixtures. Other suites of heavy oils and bitumens have shown comparable (i.e., among themselves) hydrogen donor abilities.49 Effect of Atmosphere on Conversion. Comparison of reactions of these five coals in model compounds (39) Moschopedis, S. E.; Hawkins, R. W.; Speight, J. G. Fuel Process. Technol. 1982, 5, 213-228. (40) Huang, L. Ph.D. Dissertation, The Pennsylvania State University, 1995. (41) Pott, A.; Broche, H.; Nedelmann, H.; Schmitz, H.; Scheer, W. Glueckauf 1933, 69, 903. (42) DeMarco, I.; Chomon, M. J.; Legarreta, J. A.; Torres, A. Fuel Sci. Technol. Int. 1991, 9, 1123-1135. (43) Nagaishi, H.; Idogawa, K.; Sasaki, M.; Maekawa, Y.; Sanada, Y.; Chiba, T. Nippon Enerugi Gakkaishi 1992, 71, 264-271. (44) Inukai, Y. Fuel Process. Technol. 1995, 43, 157. (45) Bolat, E.; Kavlak, C¸ .; Yalin, G.; Dinc¸ er, S. Fuel Process. Technol. 1992, 31, 55. (46) Ternan, M.; Rahimi, P.; Liu, D.; Clugston, D. M. Energy Fuels 1995, 9, 1011. (47) Rahimi, P. M.; Fouda, S. A.; Kelly, J. F.; Malhotra, R.; McMillen, D. F. Fuel 1989, 68, 422-429. (48) Gao, J.; Wu, Y.; Oelert, H. H.; Zhang, P. Huadong Huagong Xueyuan Xuebao 1993, 19, 561-567. (49) Rahimi, P.; Dawson, W. H.; Kelly, J. F. Fuel 1991, 70, 95.
Coal/Petroleum Residuum Interactions
[eicosane, 1-phenyldodecane, 1,4-diisopropylbenzene, 1,2,4,5-tetramethylbenzene (durene), and pyrene] showed two effects:21 There is no clear-cut advantage of using H2, in terms of increased conversion in H2 relative to that obtained in N2, for a given coal and temperature, except for a slight (but not unequivocal) benefit of H2 at 450 °C. Most often, the increased conversion obtained by using H2, relative to N2, for a given coalsolvent-temperature condition was less than that observed in the comparable coal blank experiments. Our present work using residua instead of model compounds shows essentially the same things. With residua, the average of the differences between conversion in H2 and in N2 for all experiments at 450 °C was 4.5%. The comparable value for experiments with model compounds at 450 °C was 2.0%, which was within experimental error ((2.5%). In 15 of 44 experiments with residua, the increase in conversion in H2 relative to N2 was greater in a coprocessing experiment (i.e., with residuum present) than in a blank experiment. The comparable statistic with model compounds was 19 of 75. Thus, there is a discernible, albeit rather slight, improvement in hydrogen utilization with actual residua rather than model compounds, but the effect is not so significant as to represent a major departure from previous results and their interpretation. Effect of Different Residua. The most dramatic difference observed between the coprocessing reactions with residua and the previous model compound studies is the remarkably low, and often negative, conversions observed at 450 °C with residua but not with the model compounds. The large reduction in conversions at 450 °C relative to 400 °C we attribute to retrogressive reactions, which will be discussed below. With model compounds, we found that pyrene was generally the most effective of the five in providing high conversions, regardless of temperature, atmosphere, or coal.21 At 350 °C, conversions of a given coal in residua tend to intermediate in value between those with pyrene and with the three alkylated benzenes, but closer to pyrene than to the alkylated benzenes. Using PSOC 1501 in N2 as a specific example, conversion in pyrene at 350 °C was 29%, conversions in the residua were 13-16%, and conversions in the three alkylated benzenes were 6-8%. For reactions at 400 °C, conversions in residua were comparable to those in pyrene. As an example, with PSOC 1498 in H2, the conversion in pyrene was 28%, while those in the three residua were 24-30%. Roughly, then, the residua are “pyrene-like” in their behavior as solvents at temperatures e400 °C. We had also observed that the effectiveness of the model compounds as solvents appeared to be related to their aromaticities. We have not seen the same distinction with the residua, but it must be noted that aromaticities and H/C ratios of the three residua vary across a much narrower range than did the comparable values of the five model compounds. Significant differences in conversions are observed at 450 °C with respect to the residuum used. Greater effect of petroleum feedstock on coal conversion during coprocessing is noticeable at temperatures >400 °C.19,23 The conversions with Blend and West Texas vacuum residua are mostly negative and only in a few cases slightly positive. As seen in Table 5, conversions at 450 °C with these two residua range from -20 to 3.6 wt %.
Energy & Fuels, Vol. 11, No. 1, 1997 121
Figure 1. Coal conversion for coal PSOC 1498 with three petroleum residua and hydrogen for a reaction time 30 min.
Negative conversions have been reported for coprocessing experiments conducted at 450 and 470 °C15,17-19,23 and are indicative of retrogressive reactions. (Observed negative conversions may also include transformations of the petroleum residuum, such as loss of its solvating components and a decrease of its solvating ability.) A comparison of the conversion with different residua at the three different temperatures is shown in Figure 1 for PSOC 1488 coal; a similar relationship holds for the other coals. Conversions for the coprocessing experiments with Hondo residuum are higher than those with the other two residua, and positive values for some of the coals are observed. On the other hand, conversions for a selected coal are invariably lowest in coprocessing reactions with West Texas residuum. West Texas residuum contains the highest amount of asphaltenes (Tables 2 and 3) of the samples used in this study and the highest proportion of aromatic carbon. Asphaltenes are thought to be key agents responsible for processing problems of petroleum feedstocks.22 The reactivity of West Texas residuum at 450 °C leading to greater negative coal conversions may be determined by the higher concentration of asphaltenes and relatively higher aromatic nature of this residuum. Some influence of the asphaltene content and structure has been observed in coal coprocessing reactions4,19 as well as in carbonization studies of petroleum feedstocks.50 The differences in coal conversion depending on the residuum used are quite obvious at 450 °C. Evidently, significant changes in the chemical behavior of the residuum take place as the reaction temperature is increased from 400 to 450 °C. The chemical nature of the residuum becomes an important factor at temperatures >400 °C. This is in agreement with the behavior of the residua when reacted alone, as discussed above. Below 400 °C the conversions are independent of the nature of the residuum. The petroleum residua in this temperature range have the function of a solvent for the coal particles. Compared to pure compound solvents the petroleum residua are very effective or even more effective than a nondonor solvent.21 Since it is likely (50) Eser, S.; Jenkins, R. G. Carbon 1989, 27, 889-897.
122 Energy & Fuels, Vol. 11, No. 1, 1997
that little vaporization of the residua occurs under these reaction conditions, the effectiveness of the residua may derive in part from an improved fluidity in the reacting system.51 The major differences are, however, the negative coal conversions obtained in combination with the petroleum residua at 450 °C, which were not observed when pure solvents were used. The solids are usually a result of the reactive species recombining and cross-linking before they have been stabilized by hydrogen. In coprocessing the reactive species are generated both by the coal and the residuum. Effect of Coal Type on Conversions. In these reactions, the effect of coal type is not of great importance in determining conversions. In hydrogen, at least one of the three residua will provide conversions g30% for each coal. As with model compound reactions,21 conversions at 350 °C are invariably lower than those at 400 °C for all coals. At 400 °C, conversions with the caking coals (PSOC 1501, 1504, and 1448) are, as a group, higher than those of the noncaking coals. We explained this on the basis of the earlier work of Marzec and Schulten.52 The highest conversions at 400 °C are obtained with the coals of highest free swelling index (FSI). Our previous work has shown that two types of behavior can be observed, which are related to the caking behavior of the coals.21,23 In liquefaction studies the caking coals (or fusible coals) have been found to be more reactive than noncaking (nonfusible) coals.21,23,53,54 Influence of Time on the Coal/Residuum Interactions. As remarked above, one of the significant distinguishing features between these reactions with actual residua and reactions of the same coals with model compounds under the same atmosphere and temperature is the very low, and often negative, conversions at 450 °C in coprocessing. The overall negative conversions measured at 450 °C may be a result of the coking reactions of each feedstock independently or the product of interactions between the two components. The conversion values alone cannot discriminate between the solids originating from coal or from residuum. Carbon isotope studies indicate that the coal contributions to the different product fractions increase with the increase in boiling point of the fraction and that a greater portion of the insoluble fraction originates from coal.55-57 In 25 of the 29 reactions the weight of THFinsolubles recovered is greater than would be accounted for by a simple sum of the THF-insolubles from the appropriate coal and residuum blanks. (The four exceptions all occurred with West Texas residuum in N2.) This strongly suggests that the low conversions are a result of chemical interactions between the residuum and the coal. The major difference between the model compounds and the residua is that the latter experience some decomposition at 450 °C, as indicated by the yields of THF-insolubles in the blank experiments. In these (51) Dhawan, J. C.; Legendre, R. C.; Posey, S. M. Fuel 1991, 70, 30-37. (52) Marzec, A.; Schulten, H. R. Prepr. Pap.sAm. Chem. Soc., Div. Fuel Chem. 1991, 36, 454-461. (53) Mochida, I.; Takerabe, A.; Takeshita, K. Fuel 1979, 58, 17-23. (54) Mochida, I.; Iwamoto, K.; Tahara, T.; Korai, Y.; Fujitsu, H.; Takeshita, K. Fuel 1982, 61, 603-609. (55) Steer, J. G.; Ohuchi, T.; Muehlenbachs, K. Fuel Process. Technol. 1987, 15, 429-438. (56) Bottrell, S. H.; Bartle, K. D.; Louie, P. K. K.; Tylor, N.; Kemp, W.; Steedman, W.; Wallace, S. Fuel 1991, 70, 442-446. (57) Keogh, R. A.; Hardy, R. H.; Davis, B. H. Energy Fuels 1991, 5, 322-327.
Tomic` and Schobert
Figure 2. Influence of reaction time on the yield of THFinsolubles from noncaking coal PSOC 1488 (coal “blank”) in hydrogen.
Figure 3. Influence of reaction time on the yield of THFinsolubles from caking coal PSOC 1504 (coal “blank”) in hydrogen.
systems, retrogressive reactions arise when both the coal and the residuum are undergoing active thermal decomposition. To further investigate this point, we studied the coal/ residuum interactions with respect to time. Two coals and two residua were reacted independently under H2 at several reaction times from 15 to 60 min. The yield of THF-insoluble matter from each coprocessing component was compared to the THF-insolubles from the coprocessing reaction for the given coal/residuum pair. One coal was selected as representative of noncaking coals (PSOC 1488) and one as representative of caking coals (PSOC 1504). The influence of reaction time on the yield of THF-insoluble matter at 350, 400, and 450 °C for the two selected coal samples is shown in Figures 2 and 3. Both increasing temperature and increasing time reduce the yield of insoluble matter when a subbituminous, noncaking coal (PSOC 1488) is reacted without a solvent (Figure 2). The yield of insoluble matter from PSOC 1488 decreases with increase of reaction time for all three temperatures. A decrease of about 15% is observed with the increase of temperature at any given reaction time. For this coal, the temperature seems to have a larger effect than the reaction time in terms of improving conversion to THF-soluble products. The influence of reaction time and temperature on the yield of insoluble matter for the selected caking coal, PSOC 1504, is shown in Figure 3. The first noticeable difference relative to the previous case is that the yield of THF-insolubles does not decrease steadily with increasing temperature. While the temperature increase from 350 to 400 °C reduces the yield of insolubles,
Coal/Petroleum Residuum Interactions
Energy & Fuels, Vol. 11, No. 1, 1997 123
Figure 4. Influence of reaction time on the yield of THFinsolubles from petroleum residua at three temperatures in hydrogen.
Figure 6. Effect of reaction time on the amount of THFinsolubles for Hondo residuum, two coals, and the coprocessing reactions at 450 °C in hydrogen.
Figure 5. Effect of reaction time on the amount of THFinsolubles for Hondo residuum, two coals, and the coprocessing reactions at 400 °C in hydrogen.
Figure 7. Effect of reaction time on the amount of THFinsolubles for blend residuum, two coals, and the coprocessing reactions at 450 °C in hydrogen.
further increase from 400 to 450 °C actually causes an effect in the opposite direction. This was the case for all reaction times, confirming the reports that the caking coal has an optimum temperature in terms of conversion to THF-soluble products.21,23 For both coals, the effect of reaction time is much weaker than the effect of temperature. Also, in Figures 2 and 3 the point corresponding to unreacted coal is 100 wt % on the abscissa and, thus, the results indicate that much of the conversion occurs within the first 15 min of reaction. Hondo and Blend petroleum residua were also tested at times up to 60 min. Figure 4 shows that the insolubles yields are insignificant at 350 and 400 °C, even at the longest reaction time, and that the induction times are relatively long. The induction time for Hondo residuum is greater than the longest time used in these experiments, while for Blend residuum the induction time is ≈40 min at 400 °C. On the other hand, the slopes of the curves at 450 °C indicate a sharp increase in yield of THF-insolubles with time. The maximum observed yield for Hondo residuum is 10 wt % and for Blend is 14.5 wt %. The amount of insolubles (in grams) from coprocessing pairs is compared with the amount of insolubles from each individual feedstock at the same reaction conditions and presented in Figures 5-7. Although the two ordinates have different absolute values, it is important to note that the slopes of the lines are directly comparable as the scales are of the same increments. The solid lines in these figures denote the variation of insoluble matter from the residuum and the coprocessing reactions, and the dashed lines, from the coals.
Effect of reaction time on the amount THF-insoluble from coprocessing Hondo residuum with PSOC 1488 and PSOC 1504 at 400 °C is shown in Figure 5. Very similar figures were obtained for reactions with Blend residuum and for reactions at 350 °C. The position of the lines denoting the insoluble product from coprocessing indicate a decrease of insolubles compared to coal “blank”. These results at various reaction times confirm the earlier results at 30 min, that the residuum improves the coal conversion to THF-soluble products at 350 and 400 °C. At these temperatures, the residuum is a relatively good solvent, helping the coal fragments to be converted to soluble products. The only difference between Hondo and Blend residuum in coprocessing reactions is the response to longer reaction times, when the latter starts producing an increased amount of insolubles, corresponding to the induction time observed in Figure 4. At reaction times 400 °C. The degree of retrogressive reactions is influenced by the degree of polycondensed structures in the residua as well as the caking behavior of the coal. Specifically, in this temperature range the results show that the residua with a lower degree of polycondensation (Har/ Car > 0.5) are preferred for inhibiting retrogressive reactions. On the other hand, a residuum with higher degree of polycondensation (West Texas, Har/Car ) 0.4)
Figure 8. Incremental improvements in conversions obtained by adding a nondonor or nonshuttler liquid medium to the reactor, then using petroleum residua, and finally using pyrene with hydrogen. Use of residua above 400 °C causes marked falloff in conversions due to retrogressive reactions.
and a noncaking coal (FSI e 0.5) proved to be the poorest combination. The rates of formation of insolubles during coprocessing depend on the characteristics of the feedstock and their individual rates of solid formation. Our results indicate that the solids from coprocessing reactions are generated at a much greater rate than from each individual feedstock and are not a simple sum of the coal and residuum solids. These results provide evidence of coal/petroleum residua interactions contributing to the insolubles. The total amount of insolubles in coprocessing also depends on the time of reaction and generally increases with time. While the optimum reaction time depends on the specific type of feedstock, a good indication of the optimum reaction time is the induction period of the petroleum residua. Even when reactions were carried out at temperatures >400 °C, the degree of retrogressive reactions can be minimized by controlling the time of reaction to be shorter than the induction period of the residuum sample. It appears, then, that the low or negative conversions at 450 °C arise from two factors: that both the coal and the residuum are actively undergoing thermal decomposition and that hydrogen utilization (in the absence of a catalyst) is ineffective. In a previous paper we presented a roadmap (ref 21, Figure 4) showing sequentially how choices of temperature, atmosphere, and solvent can steadily enhance conversions from 11-24% to 35-43%.21 Retrogressive reactions represent a serious “detour” from this pathway. Figure 8 is an updated and expanded version of
Coal/Petroleum Residuum Interactions
our earlier figure, now incorporating the effects of retrogressive reactions. Under the conditions we have been investigating, with no hydrogenation catalyst, no donor solvent, and low solvent/coal ratios, the best conversions can be obtained at the highest temperature (450 °C) in hydrogen and using a hydrogen shuttler solvent (pyrene). In addition, a small, but perceptible, advantage can be obtained by using caking, rather than noncaking, coals. Even though the petroleum residua appear to be “pyrene-like” in their behavior at 400 °C, pushing these liquids above that temperature, when they themselves begin to thermally decompose, causes conversions to plummet precipitously.
Energy & Fuels, Vol. 11, No. 1, 1997 125
Acknowledgment. We are pleased to thank the U.S. Department of Energy for financial support for this work. We enjoyed many useful discussions with Drs. Bruce Utz and Karl Schroeder of the Pittsburgh Energy Technology Center during the course of the work. We also thank John W. Green, Nottingham-Trent University, for experimental assistance with some of the reaction time studies and Ronald Copenhaver for assistance with the microautoclaves. The coals and their accompanying analytical data were kindly provided by the Penn State Coal Sample Bank and Data Base. EF960103W